Diffusion of Innovation

It has long been speculated that influential individuals were critical in the widespread of adoption of innovations. This view has recently been attacked by some computational network-based models by Watts and Dodds who have shown that it is not highly connected individuals that are important but instead a critical mass of easily influenced individuals. Which of these views is correct has yet to be shown, but agent-based modeling combined with social network analysis can provide a new insight into this problem. For instance, Stonedahl, Rand, and Wilensky have shown that the topology of interaction between individuals interested in adopting an innovation is critical to the speed of innovation adoption. This work has also made use of machine learning techniques that enable the agents to adapt to innovations. This enables the construction of models in which agents are not just passive conduits of information, but can manipulate the information presented them before passing the information on. By applying these methods of complexity (agent-based modeling, social network analysis, and machine learning) the project will develop a richer insight into the diffusion of innovation.